Turnitin works by comparing the text of your submission against a huge database of web pages, academic publications and previously submitted student papers, then producing a Similarity Report that highlights any matching passages and gives an overall similarity score. It is a text-matching service, not a judge: it shows where your wording overlaps with existing sources, but a human marker decides whether that overlap is acceptable referencing or actual plagiarism. This guide explains what Turnitin is, exactly how the matching engine works, how to read the Similarity Report and score, what the software does and does not do, whether it stores your paper, how its AI-writing indicator fits in, and the common myths that cause students unnecessary panic.
What is Turnitin?
Turnitin is an originality-checking and academic-integrity service used by the vast majority of UK universities and colleges, as well as thousands of institutions worldwide. When your tutor sets up an assignment to be submitted “through Turnitin”, your work is uploaded to the platform, scanned, and returned with a colour-coded Similarity Report. That report shows which parts of your text match sources Turnitin already knows about.
The crucial thing to understand from the outset is what Turnitin is for. It is a tool that supports markers and promotes good academic practice. It is designed to surface unoriginal or poorly attributed material so that a human can investigate. It does not make accusations, assign penalties, or decide guilt. Knowing this changes how you read your report entirely: a number on a screen is information, not a verdict.
Turnitin is also widely misunderstood. Many students treat the similarity percentage as a pass/fail line, panic over a perfectly normal score, or assume the software “knows” they have plagiarised. None of that is how the system actually behaves. Once you understand the mechanics, the report becomes a genuinely useful proofreading aid rather than a source of dread. If you want a refresher on the underlying concept first, our explainer on what counts as plagiarism is a good companion to this guide.
How does Turnitin work: the matching engine
At its core, Turnitin performs text matching. When you submit a document, the platform breaks your writing into overlapping strings of words and compares those strings against everything in its repositories. Where a run of your text lines up closely with text it already holds, it records a match and highlights it in the report. Turnitin compares your submission against three broad collections:
- The open web — billions of current and archived web pages, including blogs, news sites, study-help pages, course notes and essay mills.
- Academic and publication databases — a large body of journal articles, books, conference papers and periodical content licensed from publishers and content partners.
- Previously submitted student papers — a private repository of work submitted by students at participating institutions, often going back many years. This is why two students who copy from the same older essay will both be flagged, even if that essay was never published online.
The third collection is the part students underestimate most. If a friend in the year above submitted an essay that is now stored in the repository, and you reuse a few sentences of it, Turnitin will match your text against their paper even though it appears nowhere on Google. Reusing your own earlier submission can be flagged too — this is self-plagiarism, and it is treated as a real integrity issue at most institutions.
It is worth being precise about a common misconception: Turnitin matches text, not ideas. It cannot tell whether you genuinely understood a concept or whether you have correctly credited an idea you paraphrased in your own words. It only sees overlapping wording. That has two consequences. First, a properly paraphrased and cited passage may show little or no match. Second, a perfectly legitimate quotation, a standard methodology sentence, or a common phrase can light up as a match even though nothing is wrong. The software flags the overlap; a person interprets it.
The Similarity Report and the similarity score
After processing, Turnitin returns the Similarity Report. This is your document with matching passages highlighted in colour, each tied to a numbered source in a side panel. The headline figure is the Similarity Index (also called the similarity score) — a single percentage representing how much of your text matched something in the databases.
So if your report reads 18%, it means roughly 18% of your wording overlaps with existing sources. That figure on its own tells you very little. A 30% score made up entirely of correctly quoted and referenced material can be completely fine, while a 9% score that turns out to be one uncited paragraph lifted from a journal is a serious problem. The percentage is a starting point for investigation, not a grade.
This is exactly why the breakdown matters more than the headline number. A sensible way to read the report is to work through each highlighted match and ask what it actually is. Universities rarely publish a single “safe” threshold, because the right number depends on the subject, the assignment type and how much quoting is appropriate. There is no universal pass mark, so treat any rule of thumb you hear with caution and check your own department’s guidance on an acceptable Turnitin similarity score.
What Turnitin flags versus what it does not do
Drawing a clear line between the software’s job and the marker’s job removes most of the anxiety around Turnitin. The table below sets out what the tool detects and what it deliberately leaves to a human.
| What Turnitin DOES | What Turnitin does NOT do |
|---|---|
| Matches your text against web pages, journals, books and stored student papers | Decide whether you have plagiarised — that is a human judgement |
| Highlights overlapping passages and links each to its source | Understand context, intent or whether a match is properly cited |
| Calculates an overall similarity percentage (the Similarity Index) | Set a pass/fail threshold or assign a grade |
| Flags reused or self-recycled text, even if it is offline | Distinguish a legitimate quotation from copied text on its own |
| Provides a separate AI-writing indicator (where enabled) | Prove with certainty that a human or AI wrote a passage |
| Gives markers a structured starting point for review | Replace academic judgement or institutional misconduct procedures |
The single most important row is the first on the right: Turnitin does not decide plagiarism. The phrase you will sometimes hear — “Turnitin caught me” — is misleading. Turnitin highlighted an overlap; a marker read it, considered the context, checked your referencing and reached a conclusion. The software is the smoke alarm, not the fire investigator. Understanding precisely which types of plagiarism exist helps you see why context is everything: copying, mosaic or patchwork writing, and missing citations are all treated very differently from a well-attributed quotation.
Does Turnitin store your paper?
This is one of the most common — and most reasonable — questions students ask. The honest answer is: it depends on how your institution configured the assignment.
When a tutor sets up a Turnitin assignment, they choose a repository option. The two outcomes that matter to you are:
- Stored in the standard repository — your paper is added to Turnitin’s database of student work. Future submissions (yours or anyone else’s) will be matched against it. This is the most common setting for final assessed work.
- No repository — your paper is checked but not added to the database. Tutors often use this for draft check-points so that your final submission does not match your own draft at 100%.
This setting explains a frequent panic: a student runs a draft through a Turnitin assignment, it gets stored, and then their final version comes back at 95% similarity — because it is matching against their own stored draft. If that happens, it is not a misconduct finding; it is a configuration artefact, and your tutor can disregard or reset it. It is always worth asking your tutor which repository setting an assignment uses before you submit a draft.
Because a stored submission can interfere with later checks, you should be careful about how many times you push the same work through an official Turnitin dropbox. This is also a strong argument for pre-checking your originality through a separate, non-storing tool first, which we cover below.
Turnitin’s AI-writing indicator
Alongside the Similarity Report, many institutions now have access to Turnitin’s AI-writing indicator, which attempts to estimate how much of a document may have been generated by AI tools such as large language models. It is important to be clear that this is a separate feature from similarity matching: text written by AI is often entirely original, so it would not show up as a similarity match at all. The AI indicator looks at writing patterns rather than overlap with existing sources.
The AI indicator is also less mature and more contested than text matching. It produces an estimate, not a definitive ruling, and — like the similarity score — it is intended to inform a human reviewer, not to act as a verdict. False positives and false negatives are both possible, which is exactly why responsible institutions treat the indicator as one signal among several. We cover the mechanics, accuracy debates and what an AI flag actually means in our dedicated guide on whether Turnitin detects AI writing. If you want to sense-check your own work for AI-style patterns before submitting, our AI content detector gives you a private read on it.
“Our Similarity Report is not a plagiarism report. It is a tool to help instructors and students identify text that may require closer review.” — Turnitin guidance for students
Common Turnitin myths, debunked
A lot of stress around Turnitin comes from folklore passed between students. Here are the persistent myths and the reality.
- Myth: any similarity score means you cheated. Reality: references, quotations and common phrases all generate matches. A non-zero score is normal and expected.
- Myth: there is a magic safe percentage (often quoted as 10% or 15%). Reality: there is no universal threshold. What matters is what the matches consist of, and your department’s own guidance.
- Myth: Turnitin decides whether you are guilty. Reality: it only highlights matching text. A human marker investigates and decides.
- Myth: swapping a few words or using synonyms beats the system. Reality: superficial word-swapping is itself poor practice and can still match. Genuine paraphrasing means rewriting in your own words from your own understanding, then citing the source. The goal is to write well and credit properly, never to evade the tool.
- Myth: Turnitin reads images, so I can submit text as a picture. Reality: doing this to hide content is academic misconduct in itself, and institutions explicitly prohibit it.
- Myth: a quotation will always be flagged as plagiarism. Reality: a correctly quoted and cited passage may show as a match, but a marker recognises it as legitimate attribution, not plagiarism.
The throughline is simple: Turnitin rewards honest writing and accurate referencing. The legitimate way to lower your similarity score is to quote less and paraphrase more, attribute everything properly, and remove any text that is too close to a source. If you find matches you need to address, our walkthrough on how to remove plagiarism covers the proper, integrity-first methods, and a good paraphrasing tool can help you rework awkward sentences while you keep the citation in place.
Why a pre-check before submission helps
Here is an honest practical point that universities rarely spell out: in most cases you only get one official Turnitin check — the one that happens when you submit your final assignment. By then it is too late to act on what the report tells you. Once it is submitted, it is marked.
That asymmetry is why pre-checking your work through a separate tool is so valuable. It lets you see the overlap, fix any genuine issues, tidy your referencing and quote correctly before the submission that counts. To be clear, this is not about gaming anything — it is about catching honest mistakes (a missing citation, an over-close paraphrase) while you can still correct them properly.
You have two sensible options:
- A quick scan. Our free plagiarism checker runs a basic web-based originality check (up to around 3,000 words) so you can spot obvious copied passages and unattributed text fast, at no cost.
- A full, Turnitin-level report. For the complete picture — matching against academic databases and the wider repository, plus AI-writing detection — our detailed plagiarism report gives you a thorough breakdown that mirrors what your marker will see, so there are no surprises on submission day.
Whichever you choose, the workflow is the same: check early, read the matched passages rather than fixating on the headline number, fix any real problems by paraphrasing and citing correctly, and only then submit. Getting your referencing right — whether Harvard or another style — does more to keep your score healthy than anything else, because accurate citation is what turns a “match” into legitimate scholarship.
See your similarity before your tutor does
Get a full Turnitin-level report with database matching and AI-writing detection — and fix issues before you submit.
Key takeaways
Turnitin is a text-matching service, not an automated judge. It compares your writing against the web, publication databases and a repository of student papers, then produces a Similarity Report and an overall percentage. That percentage is information to be interpreted, never a grade in itself — what matters is what the highlighted matches actually are. The software flags overlap; a human marker decides whether anything is wrong, taking your citations, quotations and context into account. Its AI-writing indicator is a separate, evolving signal that also informs rather than rules. Because you usually get only one official check, pre-checking your work and referencing accurately are the two most reliable ways to walk into submission day with confidence.
